DE202014010689U1 - A system for evaluating the quality of medical images of at least a portion of a patient's anatomy - Google Patents

Abstract

A system for assessing the quality of medical images of at least a portion of a patient's anatomy, the system comprising: a digital storage device storing instructions for assessing the quality of medical images of at least a portion of a patient's anatomy, and a processing unit configured thereto is to carry out the instructions to perform a method comprising: receiving one or more images of at least a portion of a patient's anatomy; performing anatomical localization or modeling of at least a portion of the patient's anatomy based on the received images; Obtained for a selected one of the received images of a plurality of image characteristics of the selected image of the received images by using the processing unit of the computer system, wherein the plurality of image characteristics comprise two or more different types of image characteristics and wherein each of the plurality of image characteristics is associated with an anatomical peculiarity of the anatomy of the patient based on the anatomical location or modeling, and calculating an image quality score for the selected image of the received images based on the plurality of image characteristics of the selected image of the received images using Processing unit of the computer system.

Description

 PRIORITY

This application claims the benefit of priority of US Provisional Application No. 61 / 793,162, filed Mar. 15, 2013, which is incorporated herein by reference in its entirety.

 TECHNICAL FIELD OF THE INVENTION

Embodiments of the present disclosure relate to medical image quality assessment systems, and more particularly, to medical image quality assessment systems relating to patient-specific anatomy and / or blood flow modeling.

 BACKGROUND OF THE INVENTION

Medical imaging is an important technology used to capture anatomical and physiological data about the body, organs, tissues of a patient, or part of them, for clinical diagnosis and planning of a treatment. Medical imaging includes, but is not limited to, radiography, computed tomography (CT), magnetic resonance imaging (MRI), fluoroscopy, single-photon emission computed tomography (SPECT), positron emission tomography (PET), scintigraphy, ultrasound and specific techniques such as echocardiography, mammography, intravascular ultrasound and angiography. Image data can be acquired by noninvasive or invasive procedures. The fields of cardiology, neuroscience, oncology, orthopedics, and many others can benefit from information gathered in medical imaging.

Particularly in the field of cardiology, it is well known that diseases of the coronary arteries can lead to the development of lesions on the blood vessels leading blood to the heart, such as a stenosis (abnormal constriction of a blood vessel). This may limit blood flow to the heart. A patient suffering from coronary artery disease may experience chest pain which, when encountered during exercise, is termed chronic stable angina or, when the patient is at rest, unstable angina. A more severe manifestation of the disease can lead to a myocardial infarction or a heart attack. There is a need to provide more accurate data on coronary lesions, e.g. size, shape, location, functional significance (eg, whether the lesion affects blood flow), etc. Patients suffering from chest pain and / or having symptoms of coronary artery disease may be subjected to one or more tests, as based on medical imaging, which may provide some indirect evidence regarding coronary lesions.

In addition to CT, SPECT and PT, the use of medical imaging for non-invasive coronary evaluation may include electrocardiograms, evaluation of biomarkers from blood tests, treadmill tests and echocardiography. However, these noninvasive tests typically do not provide a direct assessment of coronary lesions or blood flow rate assessments. The non-invasive tests can provide indirect evidence of coronary lesions by following cardiac electrical activity (eg, using electrocardiography (ECG)), myocardial motion (eg, using exercise echocardiography), perfusion of the myocardium (eg PET or SPECT) or metabolic changes (eg using biomarkers).

For example, anatomical data can be acquired non-invasively using coronary computed tomography angiography (CCTA). CCTA can be used for imaging patients with chest pain and includes the use of CT technology to image the heart and coronary arteries after intravenous infusion of a contrast agent. However, CCTA also can not provide direct information about the functional significance of coronary lesions, e.g. whether the lesions affect the blood flow. In addition, since CCTA is an exclusively diagnostic test, it can not be used to provide changes in coronary blood flow, pressure, or myocardial perfusion under other physiological conditions (e.g., exercise), nor can it be used to predict the outcome of procedures.

Thus, patients may also require an invasive test, such as diagnostic cardiac catheterization, to image coronary lesions. Diagnostic cardiac catheterization may include performing conventional coronary angiography (CCA) to obtain anatomical data on Collect coronary lesions by providing a physician with a picture of the size and shape of the arteries. However, CCA does not provide data to evaluate the functional significance of coronary lesions. For example, a physician may not be able to diagnose whether a coronary lesion is detrimental without determining whether the lesion is functionally significant. Thus, CCA has led to what is referred to as an "oculostenotic reflex" by some interventional cardiologists to use a stent for any lesion found using CCA, regardless of whether the lesion is functionally significant. As a result, CCA can lead to unnecessary operations on patients, which can cause additional risks for patients and unnecessary treatment costs for patients.

During diagnostic cardiac catheterization, the functional significance of a coronary lesion can be assessed invasively by measuring the fractional flow reserve (FFR) of a detected lesion. FFR is expressed as the ratio of the average blood pressure downstream of a lesion divided by the average blood pressure upstream of the lesion, e.g. the aortic pressure under conditions of increased coronary blood flow, e.g. due to intravenous administration of adenosine. The blood pressures can be measured by inserting a pressure wire into the patient. Thus, the decision to treat a lesion based on the particular FFR may be taken after the initial cost and risk of diagnostic cardiac catheterization have already been incurred.

To fill in the gaps left by any of the medical imaging and invasive procedures described above, HeartFlow, Inc. has developed a simulation and modeling technology based on patient-specific image data. For example, but not limited to, computer-aided mechanics, computational fluid dynamics (CFD), numerical simulation, multi-scale modeling, Monte Carlo simulation, machine learning, artificial intelligence, and various other computer-aided methods to solve mathematical models. These techniques can i.a. Provide information about biomechanics, fluid mechanics, changes in anatomy and physiology over time, electrophysiology, stress and strain on the tissue, organ function and neurological function. This information may be provided at the time of the imaging study or the prediction of changes over time as a result of medical intervention or the course of time and progression of the disease.

An illustrative application of computer-aided simulation and modeling is described by HeartFlow, Inc. for modeling vascular blood flow from noninvasive image data, including evaluating the effect of various medical, interventional, or surgical treatments (see, eg, US Pat U.S. Patents No. 8,386,188 . 8,321,150 . 8,315,814 . 8,315,813 . 8,315,812 . 8,311,750 . 8,311,748 . 8,311,747 and 8,157,742 ). In particular, HeartFlow, Inc. has developed noninvasive methods for evaluating coronary anatomy, myocardial perfusion, and coronary artery flow to reduce the above-mentioned disadvantages of invasive FFR measurements. In particular, CFD simulations have proven successful which are used to predict spatial and temporal variations in flow rate and blood pressure in arteries, including FFR. Such methods and systems are useful for cardiologists who diagnose and schedule treatments for patients with suspected coronary artery disease, and predict coronary artery flow and myocardial perfusion under conditions that can not be directly measured, eg, during exercise, and for medical consequences , interventional and surgical treatments to predict coronary artery blood flow and myocardial perfusion.

For the techniques described above, as well as many other applications of imaging based modeling and simulation, the characteristics and quality of the image data are important. While obtaining medical image data, there may be multiple artifacts or limitations that may affect the quality of the image. For example, the settings and capabilities of spatial and temporal resolution, energy-tissue interactions, patient or organ movement, reconstruction algorithms, hardware errors, timing or acquisition, detector sensitivity, administered medication or contrast agents, patient preparation, and various other factors can affect the resulting image quality. These effects include, but are not limited to, poor resolution, motion or blurring artifacts, high noise, low tissue contrast, poor perfusion, partial volume effect, distortion, truncation, shading, etc. These quality issues are the performance and accuracy of the models and affect simulations based on the image data, there is a need to determine whether image quality is appropriate or to determine the effect of image quality on modeling and simulation results.

DISCLOSURE OF THE INVENTION

There is a need for systems for evaluating and quantifying medical image quality, and more particularly for systems for assessing and quantifying medical image quality related to patient-specific modeling of blood flow.

The object is achieved by a system according to claim 1. Advantageous embodiments and further developments are the subject of the dependent claims.

The system according to the invention also implies a method, for which reason embodiments of various systems according to the invention as well as corresponding methods will be described below in order to enable the user of the invention in the best possible way to benefit from the invention.

According to one embodiment, methods are disclosed for evaluating the quality of medical images of at least a portion of the anatomy of a patient using a computer system. A method thereby comprises receiving one or more images of at least a portion of the anatomy of a patient, determining one or more image features of the received image (s) using a processing unit of the computer system, performing anatomical localization, or modeling at least a portion of a patient's anatomy based on the received images using a processing unit of the computer system, obtaining an identification of one or more image property (s) associated with an anatomical particularity of the patient's anatomy based on the anatomical location or modeling, and calculating an image quality score based on the one or more a plurality of image feature (s) and the one or more image features using a processing unit of the computer system.

In another embodiment, systems are disclosed for evaluating the quality of medical images of at least a portion of a patient's anatomy. A system includes a digital storage device storing instructions for assessing the quality of medical images of at least a portion of a patient's anatomy, and a processing unit configured to execute the instructions to perform a method comprising: receiving one or more images at least a portion of the anatomy of a patient, determining one or more image features of the received image (s) using a processing unit of the computer system, performing anatomical localization or modeling at least a portion of the anatomy of a patient based on the received images using a processing unit of the computer system, Obtaining an identification of one or more image characteristics associated with an anatomical particularity of the anatomy of the patient based on the anatomical location or modeling, and calculating a B image quality score based on the one or more image feature (s) and the one or more image properties using a processing unit of the computer system.

 According to another embodiment, there is disclosed a non-transitory computer-readable medium for use on at least one computer system that includes computer-executable programming instructions for evaluating the quality of medical images of at least a portion of a patient's anatomy that, when executed by the at least one computer system, causes performance of a method. comprising receiving one or more images of at least a portion of the anatomy of a patient using a processing unit of the computer system, determining one or more image features of the received images, performing anatomical localization, or modeling at least a portion of the anatomy of a patient based on the received images using a processing unit of the computer system, obtaining an identification of one or more image characteristics associated with an anatomical feature of the patient's anatomy; based on the anatomical localization or modeling, and calculating an image quality score based on the one or more image feature (s) and the one or more image features using a processing unit of the computer system.

Additional embodiments and advantages will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the disclosure. The embodiments and advantages are realized and attained by means of the elements and combinations particularly set forth below.

Further details and advantages of the invention will become apparent from the following purely exemplary and non-limiting description of an embodiment in conjunction with the drawings comprising eleven figures.

 BRIEF DESCRIPTION OF THE DRAWING

1 FIG. 10 is a schematic diagram of a system for determining various information regarding coronary blood flow in a specific patient, according to an example embodiment.

2 FIG. 3 is a flowchart of a method for determining various information regarding coronary blood flow in a specific patient, according to an example embodiment.

3 FIG. 10 is a flowchart describing an exemplary method for evaluating medical image quality, generating a measure of image quality, and using a measure of image quality, according to various exemplary embodiments.

4 FIG. 10 is a flow chart describing an example method for enabling and performing a user-controlled medical image quality assessment, according to an example embodiment.

5 FIG. 10 is a flow chart describing an example method for performing computer-controlled evaluation of medical image quality, generating a measure of image quality, and using the measure of image quality, according to various exemplary embodiments.

6 FIG. 5 is a flow chart describing an exemplary method for evaluating medical image quality, generating a measure of image quality, and using a measure of image quality in conjunction with an estimate of fractional coronary flow reserve values, according to various exemplary embodiments.

7A FIG. 10 is an exemplary fractional flow reserve error box plot and approval or rejection based on a CT image quality overview, according to various exemplary embodiments.

7B FIG. 10 is an exemplary box plot of a fractional flow reserve error and a score based on a CT image quality overview, according to various exemplary embodiments.

8th FIG. 5 is an exemplary bar graph illustrating the comparisons between performance or accuracy of a fractional flow reserve and computed tomography based on image quality, by the number of vessels, according to various exemplary embodiments.

9 FIG. 13 is a table that illustrates an exemplary category for the evaluation of image characteristics based on lumen features of cardiovascular vessels, according to various exemplary embodiments.

10 FIG. 12 is a screenshot of an exemplary interface for displaying computed tomography quality scores versus reference power, according to various example embodiments.

 DESCRIPTION OF PREFERRED EMBODIMENTS

Reference will now be made in detail to the exemplary embodiments, examples of which are illustrated in the accompanying drawings. Wherever possible, the same reference numbers will be used throughout the drawings to refer to the same or like parts.

The present disclosure relates to the evaluation and quantification of the quality of medical images. In one embodiment, the present disclosure describes systems and methods for assessing image quality for the purpose of predicting or analyzing the accuracy and performance of medical imaging simulation and modeling. In one embodiment, a method for evaluating medical image quality includes receiving image data and possibly patient information, performing image quality assessment using computer-automated, user-controlled, or image quality Combining resources at the local and / or global level, and generating a measure of image quality that is regional (eg, for a vessel) or for an entire data set or multiple data sets. In one embodiment, a method of evaluating a medical image quality may include applying image quality measures to one or more of: (i) judging whether the image data is suitable for obtaining the desired simulation accuracy, precision, and / or performance; (ii) estimating the accuracy, precision or reliability of the simulation results; (iii) conducting the simulation or modeling techniques best suited to achieve the desired accuracy, precision and / or performance; and / or (iv) selecting, combining or Correct the best data from multiple received data to achieve the desired accuracy, precision, and / or performance.

In one embodiment, quality problems or anomalies may include, but are not limited to, low contrast, noise, motion or blur, registration error or misalignment, low resolution, partial volume effect, beam hardening, cut-off anatomy excluded from the scan, banding, scan errors, missing data, and / or not include matching contrast time recording. If these problems affect information of interest, such as the anatomy of the coronary arteries, in such a way as to affect the quality, accuracy or performance of the blood flow models and simulations, then it may be desirable to recognize and evaluate the image quality problems. Thereafter, the quality of the image data can be analyzed for its effect on the ability to extract the desired information from the patient images.

In an exemplary embodiment, the disclosed methods and systems include the use of at least one computer configured to receive patient-specific image data that includes at least a portion of the coronary blood stream. In order to model the coronary blood flow from the image data, at least some parts of the coronary artery anatomy may be measured, modeled, segmented or estimated. Additionally, at least some portions of the heart, aorta, myocardium, ventricles, valves, veins, and other structures of the heart may be measured, modeled, segmented, or estimated. Along with the anatomical representations, information about contrast values, contrast gradients, or other dimensions for image analysis can be extracted to notify the model of it.

Therefore, in such an exemplary embodiment, methods and systems for determining the image quality of patient-specific image data for purposes of blood flow modeling and simulation are disclosed. Such an embodiment may include evaluating coronary computed tomography angiography (cCTA) image data to simulate information including, but not limited to, coronary blood flow, velocity, pressure, plaque and wall stress, and fractional flow reserve (FFR). The methods and systems may also be adapted to other areas of the bloodstream, including, but not limited to, the carotid, peripheral, abdominal, renal, and cerebral circulation, as well as other modalities including, but not limited to, Include MRI, PET, SPECT, ultrasound and angiography.

Accordingly, in certain subsequent embodiments, systems and methods for assessing and quantifying image quality associated with images of the coronary blood circulation are described for exemplary purposes. More particularly, in certain embodiments, systems and methods for assessing and quantifying image quality associated with analyzing the image quality for exemplary purposes used to model a patient-specific coronary blood circulation and simulate blood flow through a patient-specific coronary blood stream are described. It should be understood, however, that the presently disclosed techniques for evaluating and quantifying image quality are nevertheless applicable to the evaluation and manipulation of medical imaging with respect to any anatomy or to any cardiovascular evaluation among any other medical diagnostic techniques.

 Exemplary cardiovascular relationship

In one embodiment, the present disclosure relates to methods and systems for assessing image quality associated with determining blood flow information in a specific patient using information non-invasively acquired by the patient. Various embodiments of such a method and system are described in detail U.S. Patent No. 8,315,812 , filed January 25, 2011, entitled "Method and System for Patient-Specific Modeling of Blood Flow", which is incorporated herein by reference in its entirety.

In some embodiments, the information determined by the method and system may relate to blood flow in the patient's coronary bloodstream. Alternatively, the particular information may relate to blood flow in other areas of the patient's bloodstream, such as the carotid, peripheral, abdominal, renal, and cerebral blood circulation. The coronary circulation includes a complex network of vessels ranging from large arteries to arterioles, capillaries, venules, veins, and so forth. The coronary circulation circulates blood to and in the heart and includes an aorta that conducts blood to several major coronary arteries (eg, the left anterior descending (LAD) artery, left circumflex (LCX) artery, right coronary artery (RCA), etc.), downstream of the aorta and main coronary arteries, which may further divide into branches of arteries or other types of vessels. Thus, the exemplary method and system may determine various information related to blood flow in the aorta, the main coronary arteries, and / or other coronary arteries or vessels downstream of the main coronary arteries. Although the aorta and coronary arteries (and branching branches thereof) are described below, the disclosed method and system may also refer to other types of vessels.

In an exemplary embodiment, the information determined by the disclosed methods and systems may include various blood flow characteristics or parameters, such as blood flow velocity, pressure (or a ratio thereof), flow rate, and FFR at various locations in the aorta, the major coronary arteries, and / or other coronary arteries or vessels downstream of the main coronary arteries include, but are not limited to. This information can be used to determine if a lesion is functionally significant and / or if the lesion is to be treated. This information can be determined using noninvasively collected information from the patient. Thus, the decision as to whether to treat a lesion or not can be made without the costs and risks associated with invasive surgery.

1 FIG. 10 illustrates aspects of a system for providing various information regarding coronary blood flow in a specific patient according to an exemplary embodiment. FIG. A three-dimensional model 10 The anatomy of the patient may be generated using data non-invasively acquired by the patient, as described in more detail below. Other patient-specific information can also be collected non-invasively. In an exemplary embodiment, the part of the anatomy of the patient may be defined by the three-dimensional model 10 at least a portion of the aorta and a proximal portion of the main coronary arteries (and branching branches thereof) associated with the aorta.

Various physiological laws or relationships 20 concerning coronary blood flow may be derived, for example, from experimental data, as described in more detail below. Using the three-dimensional anatomical model 10 and the derived physiological laws 20 can have several equations 30 , which are related to coronary blood flow, as will be described in more detail below. For example, the equations 30 determined and solved using any numerical method, eg, finite difference, finite-volume, spectral, lattice Boltzmann, particle-based, level set, finite element methods, etc. The equations 30 may be resolvable for information (eg pressure, velocity, FFR etc.) regarding the coronary blood flow in the anatomy of the patient at various points in the model 10 to determine the anatomy represented.

The equations 30 can be done using a computer system 40 be solved. Based on the solved equations, the computer system can 40 to output one or more images or simulations containing information regarding the blood flow in the model 10 show the patient's anatomy represented. For example, the image (s) may be a simulated blood pressure model 50 , a simulated blood flow or velocity model 52 , a calculated FFR (cFFR) model 54 etc., as described in more detail below. The simulated blood pressure model 50 , the simulated blood flow model 52 and the cFFR model 54 provide information regarding the respective pressure, velocity and cFFR at different locations along three dimensions in the model 10 represented anatomy of the patient. The cFFR can be thought of as the ratio of blood pressure at a particular location in the model 10 divided by the blood pressure in the aorta, eg at the influence limit of the model 10 under conditions of increased coronary blood flow, eg, conventionally caused by the intravenous administration of adenosine.

In an exemplary embodiment, the computer system 40 comprise one or more non-transitory computer-readable storage devices that store / store instructions that, when they are executed by a processing unit, a computer system, etc., may perform any of the actions described herein for providing various sources of information regarding blood flow in the patient. The computer system 40 may include a desktop or portable computer, workstation, server, personal digital assistant, or any other computer system. The computer system 40 For example, a processing unit, a read only memory (ROM), a random access memory (RAM), an input / output (I / O) adapter for connecting peripherals (eg, an input device, output device, memory device, etc.), a user interface adapter for connecting Input devices such as a keyboard, a mouse, a touch screen, a sound input and / or other devices, a communication adapter for connecting the computer system 40 with a network, a display adapter for connecting the computer system 40 with a display, etc. For example, the display can be used to model the three-dimensional 10 and / or any by solving the equations 30 to display generated images, such as the simulated blood pressure model 50 , the simulated blood flow model 52 and / or the cFFR model 54 ,

2 shows aspects of a procedure 60 for providing various sources of information regarding blood flow in a specific patient, according to another exemplary embodiment. The method may include acquiring patient-specific anatomical data, such as information regarding the anatomy of the patient (eg, at least a portion of the aorta and a proximal portion of the main coronary arteries (and branching branches thereof) connected to the aorta) and preprocessing the data (FIG. step 62 ). The patient-specific anatomical data can be recorded non-invasively, eg by CCTA.

A three-dimensional model of the patient's anatomy can be generated based on the acquired anatomical data (step 64 ). For example, the three-dimensional model may be the one described above 1 described three-dimensional model 10 be the anatomy of the patient.

The three-dimensional model can be prepared for analysis and boundary conditions can be determined (step 66 ). For example, the above may be related to 1 described three-dimensional model 10 trimmed to the anatomy of the patient and discretized into a volumetric network, eg a finite element or finite volume mesh. The volumetric network can be used to connect with the above 1 described equations 30 to create.

Boundary conditions can also be assigned and linked to the above 1 described equations 30 to get integrated. The boundary conditions provide information about the three-dimensional model 10 The limits of influence may include the boundaries through which the flow is directed into the anatomy of the three-dimensional model, such as at the end of the aorta near the aortic root. Each influence boundary may be associated by coupling a heart model and / or a lumped parameter model to the boundary, etc., eg, a predetermined value or field for velocity, flow rate, pressure or other property. The outflow boundaries may include the boundaries through which the flow is directed outward from the anatomy of the three-dimensional model, such as at one end of the aorta near the aortic arch, and at the downstream ends of the main coronary arteries and the branches branching therefrom. Each outflow boundary may be associated, eg, by coupling a lumped parameter or distributed (eg, one-dimensional wave propagation) model. The predetermined values of influent and / or outflow boundary conditions may be determined by noninvasively measuring physiological characteristics of the patient, such as cardiac output (the volume of blood flow from the heart), blood pressure, myocardial mass, etc., but this listing is not exhaustive. The vessel wall boundaries may be the physical boundaries of the aorta, the major coronary arteries and / or other coronary arteries or vessels of the three-dimensional model 10 include.

The computational analysis can be performed using the prepared three-dimensional model and the determined boundary conditions (step 68 ) to determine blood flow information for the patient. For example, the computational analysis with the equations 30 and using the above in conjunction with 1 described computer system 40 to be done in conjunction with the above 1 described images, such as the simulated blood pressure model 50 , the simulated blood flow model 52 and / or the cFFR model 54 ,

The method may also include providing patient-specific treatment options using the results (step 70 ). For example, in step 64 generated three-dimensional model 10 and / or in step 66 assigned boundary conditions to modeling one or more treatments, eg, inserting a coronary stent into one in the three-dimensional model 10 represented coronary arteries or other treatment options. Thereafter, the computational analysis may be as above in step 68 described to generate new images, such as updated versions of the blood pressure model 50 , the blood flow model 52 and / or the cFFR model 54 , These new images can be used to determine a change in blood flow velocity and pressure as the treatment option (s) are taken over.

The systems and methods disclosed herein may be integrated into a software tool accessed by physicians to provide a noninvasive means of quantifying blood flow in the coronary arteries and to assess the functional significance of coronary artery disease. In addition, physicians may use the software tool to predict the impact of medical, interventional, and / or surgical treatments on coronary artery blood flow. The software tool can diagnose the condition in other parts of the cardiovascular system, including the arteries in the neck (eg carotid arteries), the arteries in the head (eg, cerebral veins), the arteries in the thorax, the arteries in the abdomen (eg, the abdominal aorta and its branches) prevent, diagnose, alleviate and / or treat arteries in the arms or arteries in the legs (eg, the femoral and popliteal arteries). The software tool can be interactive to allow the physician to develop optimally personalized therapies for patients.

For example, the software tool may be at least partially integrated into a computer system, for example, that in 1 illustrated computer system 40 that is used by a doctor or another user. The computer system may receive data that has been noninvasively acquired by the patient (eg, data used to model the three-dimensional model 10 to generate data used to apply boundary conditions or perform the computational analysis, etc.). For example, the data may be entered by the physician or received from another source that can access or provide such data, such as a radiology or other medical laboratory. The data may be transmitted over a network or other system for communicating the data, or directly to the computer system. The software tool can use the data to the three-dimensional model 10 or other models / networks and / or any simulations or other results to be generated and displayed by solving the above in conjunction with 1 described equations 30 be determined, such as the simulated blood pressure model 50 , the simulated blood flow model 52 and / or the cFFR model 54 , Thus, the software tool can take the steps 62 - 70 carry out. In step 70 For example, the physician may provide further inputs to the computer system to select possible treatment options, and the computer system may display new simulations to the physician based on the selected possible treatment options. Furthermore, each of the in 2 illustrated steps 62 - 70 using separate software packages or modules.

Alternatively, the software tool may be provided as part of a web-based service or other service, eg, a service provided by a separate entity from the physician. For example, the service provider may operate a web-based service and provide a web portal or other web-based application (eg, running on a server or other computer system operated by the service provider) to the physician or other users over a network or others data between computer systems. For example, the non-invasively patient-acquired data may be provided to the service provider and the service provider may use the data to describe the three-dimensional model 10 or other models / networks and / or any simulations or others by solving the above in conjunction with 1 described equations 30 produce certain results, such as the simulated blood pressure model 50 , the simulated blood flow model 52 and / or the cFFR model 54 , Then, the web-based service can provide information regarding the three-dimensional model 10 or other models / networks and / or simulations so that the three-dimensional model 10 and / or the simulations can be displayed to the physician on the physician's computer system. Thus, the web-based service can take the steps 62 - 70 and any other steps described below for providing patient-specific information. In step 70 For example, the physician may provide further inputs, for example, to select possible treatment options or to adjust the computational analysis, and the inputs may be transmitted to the computer system operated by the service provider (eg via the web portal). The web-based service may generate new simulations or other results based on the selected possible treatment options and communicate information regarding the new simulations back to the physician so that the new simulations can be displayed to the physician.

Image quality evaluation

The computational modeling techniques described above for non-invasively calculated FFR may benefit from image quality evaluations. Accordingly, the present disclosure describes methods and systems for quantifying and evaluating the effects of image quality on the available data in the anatomical and mathematical models used in the simulation of blood flow characteristics. In addition, the present disclosure describes methods and systems for evaluating the uncertainty of a vessel and other anatomical models based on local and global image features and calculating reliability intervals of simulated blood flow calculations based on a predicted uncertainty.

• – Bildauflösung
• – Schichtdicke
• – Rekonstruktionskern
• – Anzahl der gescannten Schichten
• – fehlende Schichten oder fehlende Daten
• – Erfassungsphase
• – zur Zeit der Erfassung bereitgestellte Medikation
• – Herzfrequenz zur Zeit der Erfassung
• – anatomische Daten, die zwar erwünscht, jedoch nicht in den Bilddaten enthalten sind
• – Vorhandensein anatomischer Anomalien
• – Vorhandensein implantierter Vorrichtungen oder vorheriger Operationen
• – Kontrastniveau
• – Rauschniveau
• – Kontrast-zu-Rauschen-Verhältnis
• – Deckungsfehler oder Fehlausrichtung
• – Bewegung oder Unschärfe
• – Partieller Volumeneffekt oder Überbelichtung
• – Strahlaufhärtung
• – allgemein nicht interpretierbare oder schlecht definierte Bereiche

In an exemplary embodiment, the methods and systems may implement at least one computer configured to recognize and evaluate image quality problems. In an exemplary embodiment, coronary image data is analyzed by a combination of automated and user-controlled methods using at least one computer system. As will be described in more detail below, the disclosed methods and systems may be fully automated, fully user-controlled, or both automated and user-controlled. The disclosed methods and systems may be configured to perform an assessment that includes an assessment or quantification of one or more of the below-mentioned potential image quality problems:

• - Image resolution
• - Layer thickness
• - Reconstruction kernel
• - Number of scanned layers
• - missing layers or missing data
• - Acquisition phase
• - Medication provided at the time of collection
• - Heart rate at the time of detection
• - anatomical data that is desired, but not included in the image data
• - presence of anatomical anomalies
• - Presence of implanted devices or previous operations
• - contrast level
• - noise level
• - Contrast to noise ratio
• - coverage error or misalignment
• - movement or blur
• - Partial volume effect or overexposure
• - beam hardening
• - generally uninterpretable or poorly defined areas

In an exemplary embodiment, these problems may be recognized at a global, local, or both global and local levels. A problem at the global level may include recognizing an image quality problem based on the total image volume, and in some cases may be referred to as an "image feature." A problem at the local level may include the detection space of a particular area, e.g. around some or all of the coronary arteries, coronary plaque, along one or more vessel centerline (s), etc., and in some cases may be referred to as an "image property".

In an exemplary embodiment, the image quality determination and evaluation systems and methods may utilize a combination of automated and user-controlled quantitative and qualitative assessment of the local and global image quality issues based on the aforementioned quality issues.

Image quality problems, such as CT imaging artifacts, can arise from multiple sources, including: (i) physical-based sources such as tube (kVp, mA) and photons (fluctuation, deficiency), beam hardening (stripes, dark bands, etc.). , partial volume (overexposure), undersampling (overexposure) and gantry rotation speed, (ii) patient-based sources such as heart rate, regular rhythm (motion), metal material and BMI (beam hardening), (iii) scanner-based sources, such as calibrated detector array entities or reconstruction nuclei and methods, and / or (iv) protocol-based sources, such as administration of beta-blockers (to lower heart rate), administration of contrast agent (high concentration, flow rate, single, double -, triple phase), contrast time control, etc., ECG sync and correction, nitroglycerin (to increase vessel and increase turbidity) and left versus left + right heart turbidity.

3 FIG. 10 is a flowchart illustrating an example method. FIG 100 for evaluating medical image quality, generating image quality measures, and using image quality measures according to various exemplary embodiments. In an embodiment, the method comprises 100 receiving image data of the patient (step 102 ). In particular, according to one embodiment, step 102 implementing at least one computer system for determining image quality for simulation and modeling by receiving patient-specific data regarding the body, organs, tissue or portions thereof of the patient. For example, step 102 the capture of patient-specific data 10 on the computer system 40 or on any other computer system (which may be, but is not limited to, a computer, laptop, mobile phone, mobile tablet, DSP, cloud computer system, server farm, etc.).

The procedure 100 may include performing an automated, user-controlled or combined automated and user-controlled evaluation of a local and / or global quality of the received image data (step 104 ). For example, in an automated embodiment, a computer system may automatically determine both global quality scores of an entire image or group of images, as well as local quality scores of specific portions of a single image or portions of a pictorial anatomy of a patient. In a user-controlled embodiment, a computer system may prompt a user to determine and input global quality scores of an entire image or group of images, and to determine local quality scores of specific portions of a single image or portions of a pictorial anatomy of a patient. In certain embodiments, certain aspects of the local and / or global quality scores may be formed by any combination of automated and user-controlled scores.

The at least one computer system and method may evaluate or evaluate a single, distinct, or combination of peculiarities in image quality to produce dimensions of image quality for areas of interest or for an entire image data set (step 106 ). In particular, the at least one computer system may use the score numbers to formulate a regional or dataset image quality measure based on the assessed peculiarities of the image quality. The at least one computer system may use the results of the image quality assessment as an input to perform modeling or simulation with the patient-specific data. In addition to modeling and simulating patient-specific data, such as blood flow, image quality measures can be used as inputs to any other activity or assessment.

In one embodiment, the method 100 using generated metrics to evaluate whether the image data is apt to achieve a desired simulation accuracy (step 108 ). For example, the procedure 100 using the results of the image quality assessment to allow or reject the image data for modeling or simulation based on predetermined criteria regarding accuracy, precision, performance or other requirements. In addition, the process can 100 using the results of the image quality assessment to estimate performance measures (eg, time to complete the analysis, cost of analysis), or to make a decision based on those measures, to perform modeling or simulation with the patient-specific data using at least one computer system or not perform. For example, a computer system may calculate and display a period of time to complete an analysis based on the results of the image quality evaluation. Additionally or alternatively, the computer system may calculate and display the cost of an analysis based on the results of the image quality assessment. Additionally or alternatively, the computer system may display and / or transmit a recommendation or a need to perform or not perform modeling or simulation with the patient-specific data using at least one computer system based on the results of the image quality assessment. Any such calculated information, such as the calculated period of time to complete an analysis, the cost of an analysis, and / or performing / not performing an analysis, may be displayed to a physician, technician, or other healthcare provider, either through an electronic display and / or via an electronic display electronic network.

According to another embodiment, the method 100 use the generated measures to estimate accuracy or reliability of the simulation results (step 110 ). For example, the procedure 100 include using the results of image quality assessment, to perform a modeling or simulation and to output results with a measure of reliability (eg, errors, percentage of reliability, confidence intervals, accuracy or precision of the estimates) associated with the simulation results.

According to another embodiment, the method 100 using the metrics generated to control simulation techniques that are best suited to achieve the desired simulation accuracy (step 112 ). For example, the procedure 100 using the results of image quality assessment to model or simulate using different techniques or algorithms in the entire image data set or in the relevant affected parts depending on the image quality rating to improve desired performance, accuracy, precision or other requirements or reach.

According to another embodiment, the method 100 use the generated measures to select, combine or correct the best available data to achieve a desired simulation accuracy from multiple received options (step 114 ). For example, the procedure 100 using the results of image quality assessment to correct image quality problems prior to performing the modeling or simulation to improve or achieve desired performance, accuracy, precision or other requirements. Additionally or alternatively, the method may 100 using the results of the image quality assessment to select the data set from a plurality of available data (eg, alternating series or reconstructions) best suited for performing the modeling or simulation to improve desired performance, accuracy, precision, or other requirements to reach. Additionally or alternatively, the method may 100 using the results of the image quality assessment to combine different pieces of different image data (eg, other phases or other reconstructions or modalities) to compensate for the image quality problems and perform modeling or simulation with the patient-specific data using at least one computer system to achieve a desired one Improve performance or accuracy, precision or other requirements.

According to another embodiment, the method 100 use the generated measures to provide feedback to obtain better image quality to achieve desired accuracy (step 116 ). For example, the procedure 100 using the results of the image quality assessment to evaluate or evaluate a single, distinct or a combination of image quality peculiarities in a timeframe that allows to provide feedback to the image data providing personnel so that they could correct, repeat or update the image data to meet some predefined criteria to improve or achieve desired performance, accuracy, precision or other requirements. Using the updated or corrected image data, the at least one computer system and method may perform one or more additional iterations of the modeling or simulation.

4 FIG. 10 is a flowchart illustrating an example method. FIG 120 for performing a user-controlled image quality assessment according to an example embodiment. As in 4 In one embodiment, the method may be 120 receiving anatomical image data of the patient (step 122 ). For example, step 122 the capture of image data 10 on a computer system 40 which corresponds to any of the disclosures of the above 1 and 2 matches. The procedure 120 may further comprise determining one or more centerlines of the patient's bloodstream (step 124 ). For example, step 124 using a processing unit of a computer system 40 to automatically identify one or more center lines of the patient's bloodstream, in accordance with any of the disclosures of the above 1 and 2 matches. In one embodiment, the processing unit of a computer system 40 Add center lines to the main vessels (RCA, LAD and LCX) or any other vessel that is more than 2 mm in diameter.

The procedure 120 may further include prompting a user to enter image quality problems, image anomalies, image artifacts, or other "uninterpretable areas" along each centerline using a set of visual criteria (eg, blur, motion, image artifacts, etc.) (step 126 ). For example, a processing unit of a computer system 40 initiate the display of one or more images and centerlines, prompting a user to review and inspect the images, and input about image quality issues after finding any cover artifacts, blur, stents, unwanted contrast-to-noise ratio, motion artifacts, Overexposure artifacts, calcium, scan errors, missing slices, incomplete data and so on. For example, the processing unit of a computer system 40 Create user interface elements that a user can manipulate to indicate that the user identifies any image quality problems described herein along with certain characteristics of the location, amount, or extent of these problems. In one embodiment, either the user or the processing unit of a computer system 40 characterize any region not to be interpreted either as short (eg 0-5 mm) or long (eg over 5 mm).

In one embodiment, users may be prompted to identify the contrast period and noise as "good" when an image has high contrast, low noise, and slight right contrast, as "borderline" when an image has mediocre contrast, Noise and high right contrast, and "bad" when an image has low contrast, high noise, and high right contrast. In one embodiment, users may be prompted to identify coverage errors as "good" when an image has no registration errors that affect the lumen geometry as "borderline" when an image has registration error artifacts that are nearly perpendicular to the vessel and can be corrected , and "bad" if a picture has coincidence errors that can not be corrected or are within range of the disease, so that the lumen can not be determined. In one embodiment, users may be prompted to identify movement as "good" if the movement does not affect the lumen or plaque, as "borderline" if the image reflects that the lumen is being affected but the vessel is being interpreted and can be modeled with assumptions and "bad" if the image reflects that the interpretability of the lumen is severely affected by the motion. In one embodiment, users may be prompted to identify the overexposure as "good" if slight overexposure does not affect the interpretability of the lumen, "borderline" if a high degree of overexposure may require correction, but the image still has visibility and "bad" if a severe overexposure artifact completely obscures the lumen.

The procedure 120 may further comprise receiving or calculating a score for each area of non-interpretability based on the length of the area (step 128 ). In an exemplary embodiment, rating numbers for image quality problems - either on a qualitative scale (eg, Likert scale) or with quantitative measurements - can be determined and analyzed, how they affect or predict modeling and simulation accuracy, precision, and performance. The image quality score may have absolute perturbation criteria in which a dataset is deemed unacceptable, it may have various dimensions that are weighted, combined and weighted over a range, vessel or entire dataset, or it may have a combination of both. For example, in certain embodiments, an automatic perturbation may be triggered whenever a single or combined image quality problem (s) causes 25% or more of an artery to be imperceptible (whether due to noise, movement, overexposure, poor contrast, Cover errors, etc.).

In an exemplary embodiment, measures for either an area (e.g., vessel) or a data set may be generated by the image quality evaluation system and method based on the assessments of at least some of the image quality problems described above. In one embodiment, each region of non-interpretability may receive an evaluation number based on the length (e.g., in one embodiment: 1.5 for short and 3 for long). In one embodiment, an evaluation number to reject images of a patient (ie, a "case") may include an evaluation number of 6 for a single main vessel, an evaluation number of 8 for an entire case, and / or a so-called "problem zone" having an evaluation score of 10 has been assigned.

9 FIG. 12 is a table of an exemplary category for evaluating image characteristics based on lumen features of cardiovascular vessels according to various exemplary embodiments. FIG. In particular shows 9 an exemplary embodiment of a rating rubric for assigning scores for areas of non-interpretability or other image quality issues. For example, as in the exemplary category of 9 shown, each property may be assigned a different rating number (ie, either a combination (noise, motion, contrast), motion, misalignment, noise, overexposure, contrast, or haze) based on an extent of the area affected (eg, "complete" or " low, or "long" or "short") and based on whether the identified property: (i) completely occludes the lumen or causes missing information, thus preventing detection of the disease, (ii) prevents the determination of the exact lumen boundary, however, allows detection of the present disease (eg, shows where the minimum luminal diameter ("MLD") is located would) or (iii) prevent the determination of the exact lumenal limit and prevent the recognition of the disease. It should be noted that the rating of 9 is merely an example and that any alternative evaluation mechanisms may be considered within the scope of this disclosure. For example, the rating system may be inverted such that lower rating numbers indicate lower image quality, whereas higher rating numbers indicate higher image quality. Alternatively or additionally, the rating system may be based on an exponential, logarithmic or partial scale. Alternatively or additionally, the rating system may be generated based on a color-coded and / or letter-graded scale, wherein a color and / or letter indicates a quality level of the scored images.

In certain embodiments, image quality scores may be weighted and combined with other factors, including, but not limited to: magnitude of the effect, size of the problem, affected areas, nature of the problem (eg, noise or movement), presence / absence of disease, vessel size, Place in the heart, uncertainty of the lumen definition, combination with other problems, visual interpretation, algorithm reliability, etc. A function for ranges or records can be derived using some, all, or additional weighting factors. Such an example is shown below: Quality _Range = f (Σ _i ^vessel problem _i · magnitude · type · disease · size · ^vessel size · location · lumen uncertainty) Quality _record = f (Σ _i ^record problem _i · order of magnitude · type · disease · size · vessel size · location · luminal uncertainty)

• – Bildauflösung: Pixelgröße < 0,5 mm
• – Schichtdicke ≤ 1,0 mm
• – Anzahl gescannter Schichten ≥ 64
• – fehlende Schichten oder fehlende Daten nicht akzeptabel
• – sublinguale Nitrate müssen zum Zeitpunkt der CT-Aufnahme vorhanden sein
• – Koronararterien und Myokard müssen im Datensatz vollständig enthalten sein
• – Vorhandensein anatomischer Anomalien, wie schwerer angeborener Herzfehler, sind nicht akzeptabel
• – Vorhandensein implantierter Geräte, wie Herzschrittmacher oder vorherige Operationen, wie Bypassimplantationen, sind nicht akzeptabel

In an exemplary embodiment, limits may be defined for the following criteria and unacceptable score values may result in rejection of data for coronary blood flow modeling and simulations:

• - Image resolution: pixel size <0.5 mm
• - Layer thickness ≤ 1.0 mm
• - Number of scanned layers ≥ 64
• - missing layers or missing data unacceptable
• - Sublingual nitrates must be present at the time of CT scan
• - Coronary arteries and myocardium must be completely contained in the dataset
• - The presence of anatomical anomalies, such as severe congenital heart disease, is unacceptable
• - The presence of implanted devices, such as pacemakers or previous surgeries, such as bypass grafting, are unacceptable

• – Kontrastniveau
• – Rauschniveau
• – Deckungsfehler oder Fehlausrichtung
• – Bewegung oder Unschärfe
• – Partieller Volumeneffekt oder Überbelichtung
• – allgemein nicht interpretierbare oder schlecht definierte Bereiche

In an exemplary embodiment, the following criteria may be defined at the local level. For example, for each image quality problem, an evaluation number of an order of magnitude of the effect can be generated. Other information may be added, such as the location and size of the problem, based on:

• - contrast level
• - noise level
• - coverage error or misalignment
• - movement or blur
• - Partial volume effect or overexposure
• - generally uninterpretable or poorly defined areas

The procedure 120 may further comprise calculating and outputting a total score of the non-interpretability regions for the image as a quantitative measure of image quality (step 130 ). For example, in one embodiment, the scores for each problem, weighted by size and location, may be summed for each vessel and each case.

5 FIG. 10 is a flowchart illustrating an example method. FIG 150 for performing a computer-automated evaluation of medical image quality, producing image quality measures, and using image quality measures according to various exemplary embodiments. In one embodiment, the method 150 receiving anatomical image data of a patient and creating a vascular model of a patient's bloodstream (step 152 ). The procedure 150 further comprises, using a processing unit, determining one or more global image features (s) (step 154 ).

• – Bildauflösung: Pixelgröße < 0,5 mm
• – Schichtdicke ≤ 0,9 mm
• – Rekonstruktionskern gleichwertig zu den spezifischen Herstellerfiltern
• – Anzahl der gescannten Schichten ≥ 64
• – fehlende Schichten oder fehlende Daten nicht akzeptabel
• – Aufnahmephase > 65 % und ≤ 80 %

In an exemplary embodiment, the disclosed systems and methods may include automatically assessing quantitative information that may be extracted from the image data including, but not limited to, the image resolution, slice thickness, reconstruction kernel, number of scanned slices, missing slices, or missing data and the recording phase. The information can be extracted by analyzing the dimensions or marks in the image data (eg the DICOM header). Each of these categories may have a simple admission / rejection criterion. The following specifications serve as examples:

• - Image resolution: pixel size <0.5 mm
• - Layer thickness ≤ 0.9 mm
• - Reconstruction kernel equivalent to the specific manufacturer filters
• - Number of scanned layers ≥ 64
• - missing layers or missing data unacceptable
• - Admission phase> 65% and ≤ 80%

In an exemplary embodiment, the completeness of the resolution, layer, phase, and data may not have absolute admission / rejection criteria, but rather a range of scores that contributes to an overall image quality measure for the data set. For example, the resolution and layer thickness can be combined to obtain a voxel volume (e.g., 0.4 mm x 0.4 mm x 0.75 mm). Higher or lower resolutions may increase or decrease the overall rating of the data set.

In an exemplary embodiment, information regarding an administered medication and heart rate during an imaging study may be entered into the computer system along with the study. The computer system may permit / reject a record based on this information, e.g. Lack of sublingual nitrates may require rejection of the data set. Alternatively, the presence, absence, or dose of medication that may contribute HR or other physiological measures to the overall score, or direct the method and computer system to perform modeling and simulation using various methods. For example, the lack of sublingual nitrates may instruct the use of alternative cardiac lumen segmentation algorithms to ensure suitable vessel sizes.

In an exemplary embodiment, missing anatomical data, the presence of anatomical anomalies, and the presence of implanted devices or prior operations may be detected by a user of the computer system. The presence or absence of these issues may be added to a rating number or result in an admission / rejection decision for the record. These ratings can also be done automatically.

The procedure 150 may further comprise, using a processing unit, determining one or more centerlines of the bloodstream of a patient based on the vascular model (step 156 ). The procedure 150 may further comprise determining one or more local image features at each of a plurality of centerline locations (eg, blur, motion, contrast, etc.) (step 158 ). In an exemplary embodiment, the computer system may be configured to determine such local image features or local or global image quality by implementing a fully automatic quantitative image quality assessment based on one or more of the image quality problems described herein. For example, a processing unit of a computer system 40 one or more local image features in one of the ways discussed above with respect to the user controlled method 4 automatically determine except that the computer system 40 This can also be done automatically, such as by performing an algorithm, in some cases according to the example concepts described below.

In an exemplary embodiment, the contrast and noise levels may be assessed locally (eg, at a portion of a vessel) or globally (eg, across multiple vessels or a large representative vessel or structure). This score can be made by measuring the contrast level (eg, the mean contrast in a region of interest) and the noise level (eg, the standard deviation of the contrast in a region of interest). These measurements can also be combined to produce a signal-to-noise ratio by dividing the contrast and noise measurements. In addition, the contrast and noise measurements may also include background and surrounding tissue contrast and noise to represent the difference between the area of interest (eg, coronary artery) and the background data (eg, myocardium and epicardial fat). Alternatively, the contrast, noise and contrast-to-noise ratio can be qualitatively at a local level or global scale by evaluating the degree of noise compared to reference standards (eg 1 = bad, 2 = borderline, 3 = good). In one embodiment, a processing unit of a computer system 40 calculate the noise based on an algorithm that receives as input some CT volume data and aortic mask data (eg, from a zhf file) and that has a mean aortic Hounsfield unit ("HU") value, a noise standard deviation, a mean environment -HU value and CNR outputs. In one embodiment, a processing unit of a computer system 40 calculate the contrast differences between left and right ventricles based on an algorithm that receives as inputs some CT volumes and a myomasse (longitudinal axis and segmentation) and that outputs a mean LV HU value and average RV HU value.

In an exemplary embodiment, registration errors or misalignments may be detected by searching the data or by globally searching the data set or locally near the arteries to identify where shifts occur between adjacent images. These can be recognized by a user or by the computer system. The degree of coverage error can be determined by a distance by which the data has shifted, the size of an area that is affected by it (eg length of the vessel that is affected) or by the orientation of the affected area (eg perpendicular or parallel to the Vessel). Alternatively, the coverage error can be qualitatively evaluated on a local or global scale by evaluating the degree of coverage error compared to reference standards (eg 1 = poor, 2 = marginal, 3 = good). In one embodiment, a processing unit of a computer system 40 Calculate index layer coverage error based on an algorithm that receives a CT image, outputs the locations of a peak, and evaluates values.

In an exemplary embodiment, an artifact may be detected by movement or blurring by scanning through the global data or locally near the arteries to identify areas where the image data is out of focus or has edges (eg, the edge of a vessel has soft edges) and blurred edges). These can be detected by a user or a computer system. The degree of movement can be classified by the distance of the fuzzy data, the gradient of the image data or other quantitative means. Alternatively, the motion may be qualitatively evaluated on a local or global scale by evaluating the degree of movement as compared to reference standards (e.g., 1 = poor, 2 = border, 3 = good).

In an exemplary embodiment, partial volume or overexposure artefacts may be detected by scanning through the global data or locally near the arteries to identify areas where the image data contains peculiarities that interfere with other portions of the data. These can be detected by a user or a computer system. The degree of overexposure can be classified by the intensity, size, and / or measurement of how far it spreads into adjacent structures (e.g., how much the overexposure covers the lumen). Alternatively, the overexposure can be qualitatively evaluated on a local or global scale by evaluating the amount of overexposure compared to reference standards (e.g., 1 = poor, 2 = marginal, 3 = good).

In an exemplary embodiment, artifacts may be detected by beam hardening by scanning through the global data or locally proximate to the arteries to identify areas where the image data contains dark spots or stripes that affect other portions of the data. These can be recognized by a user or by the computer system. The degree of beam hardening may be classified as the intensity, shape, and / or measurement of how far it propagates into adjacent structures (e.g., how far the beam hardening covers the lumen). Alternatively, the beam hardening can be qualitatively evaluated on a local or global scale by evaluating the degree of beam hardening compared to reference standards (e.g., 1 = poor, 2 = marginal, 3 = good).

In an exemplary embodiment, any other general properties that affect image quality may be detected by scanning through the global data or locally near the arteries to identify areas where the image data is uninterpretable or in which the definition of the peculiarity how the lumen is bad. These can be detected by a user or a computer system. They can be quantified by the degree to which they affect luminal quality compared to adjacent areas and by the extent involved. Alternatively, the properties can be qualitatively evaluated on a local or global scale by evaluating the degree of the property as compared to reference standards (e.g., 1 = poor, 2 = marginal, 3 = good).

The procedure 150 may further include the prediction of the local uncertainty in the vascular model based on the local and global image features (step 160 ). In certain embodiments For example, machine learning, regression, and other statistical techniques may be used to derive image quality functions or models based on modeling, simulation, and performance. As described in the next section, these measures can be adjusted to meet different needs.

The procedure 150 may further include using the local uncertainty prediction to calculate a confidence interval of a simulated blood flow calculation (step 162 ). The procedure 150 may further comprise calculating and outputting a total uncertainty for the vascular model as a quantitative measure of image quality (step 164 ).

Exemplary Embodiments of the Step (Step 162 ) of using the local uncertainty prediction to calculate a confidence interval of a simulated blood flow calculation will be described in more detail below. In an exemplary embodiment, the measures may be adjusted and have various correlations or related criteria to achieve purposes including, but not limited to: evaluating the sufficiency of data for automated modeling, evaluating the sufficiency of data for user-controlled interpretation, and the like / or modeling, instructing the method or system used to model data, admitting / rejecting image data, determining which of several received data is best suited for modeling (eg, alternating phases or reconstructions), providing feedback on image data obtaining improved or corrected data, tagging the results differently depending on the image quality rating numbers, and providing a reliability estimate depending on the uncertainty associated with the image quality issues. All of these purposes are related to measuring and predicting the accuracy, precision, and performance of a simulation and modeling.

In an exemplary embodiment, criteria relating the image quality measure and an error of the FFR simulation result against a reference standard of the measured FFR may be derived. Coronary vessels and / or complete datasets meeting the criteria may be allowed for processing to ensure a certain level of accuracy and precision of resolution. Vessels or complete data sets that do not meet the criteria may be correlated with a higher error rate than desired. Alternatively, vessels or complete data sets that do not meet the criteria may be forwarded to other methods and / or systems that can achieve greater accuracy.

In an exemplary embodiment, criteria may be derived that relate the image quality measure and variability in the simulated FFR results based on different users. Coronary vessels and / or complete datasets meeting the criteria may be allowed for processing to ensure a certain level of resolution precision. Vessels or complete data sets that do not meet the criteria may be correlated with higher variability than desired. Alternatively, vessels or complete data sets that do not meet the criteria may be passed on to other processes and / or systems that can achieve higher precision.

In an exemplary embodiment, criteria may be derived that relate the image quality measure and power efficiency of blood flow modeling and simulation. Records above certain rating numbers may be rejected, associated with a special processing fee, or forwarded to various resources to achieve a desired efficiency. Alternatively, based on the image quality score, an estimate of the simulation cost and / or time may be provided. For example, as image quality gets worse and worse, it may be possible to estimate higher costs or a higher price associated with manually identifying and correcting image quality characteristics or anatomical properties.

In an exemplary embodiment, criteria may be derived to highlight results from FFR simulations performed on coronary vessels and / or complete datasets that contain an area of poor image quality that does not meet the criteria. Such markers may serve to provide an indication that there is uncertainty in the resolution of the model in this domain and / or to explain what is modeled in the face of uncertainty (eg, an assumption).

In one exemplary embodiment, criteria may be derived to mark regions in models that require special processing to ensure accuracy (e.g., inspection, different algorithm, expert opinion).

In an exemplary embodiment, criteria may be set to use certain methods of determining vessel size in the presence of certain image quality problems. For example, if there is an overexposure artifact around the calcified plaque, the methods and systems for determining the lumen boundaries (and then the blood flow) in the vicinity of the artifact may differ from those in the absence of such artifact.

In an exemplary embodiment, criteria may be derived to evaluate the uncertainty or reliability of FFR simulation results performed on coronary vessels and / or complete datasets that contain an area of poor image quality that does not meet the criteria. The uncertainty or reliability may cause the FFR results to be reported as having a percentage of reliability or a confidence interval based on the image quality effect (eg, FFR is 0.87 +/- 0.05 or FFR is <0, 80 with 76% reliability).

In an exemplary embodiment, criteria may be derived to relate the image quality measure and error of FFR simulation results to a reference standard of the measured FFR.

Coronary vessels and / or complete datasets can be ranked against these criteria based on their scores, to determine which of several data would be best suited to simulate FFR scores and achieve the highest accuracy.

In an example embodiment, at least one computer system may be located at or quickly accessible from the location where the image data is generated. Criteria may be set to evaluate the image quality as it relates to the influence or prediction of FFR simulation results. Coronary vessels and / or complete datasets may be ranked against their rating scores based on their rating numbers to provide immediate feedback such that the location that generates the image data may correct or update the data until they meet the criteria, which are required to achieve a desired accuracy. Alternatively, immediate feedback could be provided with an estimate or reliability associated with reduced accuracy, thereby allowing the location that generates the image data to allow for lower accuracy when there is a clinical benefit.

6 FIG. 10 is a flowchart illustrating an example method. FIG 200 for evaluating medical image quality, generating image quality measures, and using image quality measures in the context of estimating coronary fractional flow reserve values according to various exemplary embodiments. For example, shows 6 a method for estimating coronary fractional flow reserve (FFR) values based on certain image quality assessment techniques disclosed herein, U.S. Patent No. 8,315,812 described FFR calculation techniques.

As in 6 shown, the procedure can 200 by selecting a specific patient (step 202 ) and begin by receiving the patient's image and physiological data (step 204 ). The procedure 200 may include validating the received images and data with known patient information (step 206 ), eg for user identity or for privacy reasons. The procedure 200 may include determining if the received data is acceptable (step 208 ). For example, step 208 either the admission or rejection of each of one or more received images based on any or a combination of the image evaluation and evaluation techniques disclosed herein. If any or more images are rejected, the method may 200 include generating a reject report (step 210 ). For example, the method may include capturing images that are rejected and providing feedback on the rejected images to the user to help the user capture new images that would not be rejected. For example, the computer system may send a report to the assistant, physician, or other medical personnel about the rejected images along with guidelines and / or recommendations for setting the image acquisition parameters that would allow obtaining images with higher image quality values. Such reports, rejected images, guidelines and / or recommendations may be displayed to the physician, assistants or other medical personnel via an electronic display and / or via an electronic network.

If the data is acceptable, then the procedure can 200 perform pre-processing and evaluation of the data via a radiological workstation (step 212 ). The procedure 200 can then calculate the myocardial mass (step 214 ), creating an initial tree of coronary vessels and branches (step 216 ), finding one or more vascular lobes edges (step 218 ), the pairing of limits in the main vessels with edge segmentation (step 220 ) and the detection, segmentation, removal and smoothing of plaque and calcium (step 222 ). The procedure 200 may then include determining if the segmentation using an algorithm was successful (step 224 ). If this is not the case, then the procedure can 200 include manually segmenting or correcting the segmentation (step 226 ). If the segmentation was successful, then the procedure can 200 which include determining if an independent verification of the segmentation is acceptable (step 228 ). If this is not the case, then the procedure can 200 to step 216 returning the coronary vessels and branches to the initial tree.

If the segmentation was successful (step 228 ; Yes), then the procedure can 200 output and smooth volume model (step 230 ), extracting one or more vessel centreline (s) (step 232 ), calculating vessel cross-sectional areas (step 234 ), trimming the model (step 236 ), creating a solid model (step 238 ), setting boundary conditions for hyperemia levels (step 240 ) and creating a final network (step 242 ). The procedure 200 may then include verifying that the network and the boundary conditions are acceptable (step 244 ). If this is not the case, then the procedure can 200 to step 234 return to the calculation of a cross-sectional area.

If the grid and boundary conditions are acceptable (step 244 ; Yes), then the procedure can 200 include, a hyperemia flow (step 246 ) and to verify that the resolution result is acceptable (step 248 ). If this is not the case, then the procedure can 200 refining stenosis geometry, correcting boundary conditions, etc. (step 258 ). If the resolution result is acceptable (step 248 ; Yes), then the procedure can 200 extracting a cFFR, documenting it along the vessels, and generating images for a report (step 250 ). The procedure 200 can then include the determination of whether an independent verification of the final results is acceptable (step 252 ). If this is not the case, then the procedure can 200 the return to step 216 of generating the initial tree of coronary vessels and branches.

If independent verification of the final results is acceptable (step 252 ; Yes), then the procedure can 200 completing a report (step 254 ) and forwarding the completed report to the doctor (step 256 ).

7A FIG. 10 is an exemplary box plot of a fractional flow reserve error and approval or rejection based on the CT image quality summary, according to various example embodiments. For example, the boxplot of 7A show that the rejected cases have a greater variation of FFRct error than the allowed cases.

7B FIG. 10 is an exemplary box plot of a fractional flow reserve error and a score based on the CT image quality summary, according to various example embodiments. For example, the boxplot of 7B show that a variation of an FFRct error may increase as an image quality score increases. In other words, cases with a score of between 7 and 10 have a much higher variation of the FFRct error than cases with a score of between 0 and 1.

8th FIG. 10 is an exemplary bar graph showing comparisons between fractional flow reserve quality and computed tomography based on image quality by number of vessels, according to various exemplary embodiments. FIG. In particular, the bar chart of 8th represent performance in correlation to vascular-specific quality values of CT interpretability.

10 FIG. 10 is an exemplary screenshot illustrating computed tomography quality values versus benchmark performance according to various exemplary embodiments. FIG. Like in 10 may be a processing unit of a computer system 40 provide a user interface with which a user, such as an imaging technician, physician, or other healthcare provider, can view a CT image quality score for each of multiple patients or "cases." As in one embodiment of the 10 Each case can be described as having a "good", "mediocre" or "bad" Image quality displayed. Image quality can also be compared to certain benchmark performance standards of image quality, such as an indicator in a bar chart or other type of chart. The interface may generate and display quality reports and / or suggest recommendations for improving the quality obtained. For example, if a high noise level is detected, the user interface may suggest checking the "mA / kV settings" and provide a link for guidance on noise. Of course, the user interface and policies may provide the user with an assessment of image quality and recommendations for improving image quality with respect to any image quality issues discussed herein.

The presently disclosed systems and methods may enable automatic estimation and correction of image quality problems, thereby reducing the human time and variability previously associated with a quality control overview of the image data. Furthermore, the presently disclosed systems and methods may provide a better understanding of a relationship between simulation and modeling accuracy (e.g., FFR errors) and image quality scores. Further, the presently disclosed systems and methods may enable users to better and automatically select a desired basic phase of an analytical review image, and thus better provide "problem areas" for further review or rejection of certain scans.

In one embodiment, the techniques disclosed herein may include defining the input of uncertainties, calculating FFR analysis sensitivities, and calculating confidence intervals in FFR according to any of the techniques in U.S. Application No. 13 / 864,996, filed April 17, 2013, , which is incorporated herein by reference in its entirety.

In one embodiment, the techniques disclosed herein may include performing any of the various pre-resolution techniques described in US Application No. 13 / 625,628, filed September 24, 2012, which is incorporated herein by reference in its entirety.

In one embodiment, FFR values may be detected using machine learning estimates rather than physics-based simulations. In other words, rather than performing a coronary solver, as in the '812 patent, for each of multiple collocation points, the disclosed systems and methods can efficiently estimate blood flow characteristics based on knowledge obtained by analyzing the blood flow of numerous other patients. For example, the disclosed systems and methods may include performing any of the various machine learning techniques described in US Provisional Patent Application No. 61 / 700,213, filed September 12, 2012, which is incorporated herein by reference in its entirety. Thus, in one embodiment, the FFR values may be obtained by training a machine learning algorithm to estimate FFR values for different points of a patient geometry based on the peculiarity vectors of the physiological parameters and measured blood flow characteristics of the patient, and then applying the machine learning algorithm to the Geometry and physiological parameters of a specific patient to obtain predicted FFR values.

One or more of the steps described herein may be performed by one or more human operators (eg, a cardiologist or other physician, the patient, a web-based service provider or other third-party service provider, another user, etc.) or one or more computer systems. used by one or more such human operator (s) as a desktop or portable computer, workstation, server, personal digital assistant, etc. The computer system (s) may be through a network or other method be closed to communicate data together.

Any of the aspects described above in any embodiment may be used with other embodiments described herein. Any device and device described herein may be used in any suitable medical procedure, may be passed through any suitable body lumen and body cavity, and used to image any suitable body part.

Various modifications and variations can be made to the disclosed systems and methods without departing from the scope of the disclosure. Other embodiments will be apparent to those skilled in the art from the specification and implementation of the disclosure disclosed herein. It is intended that the description and examples be exemplary only 3, the true scope and spirit of the disclosure being defined by the following claims.

QUOTES INCLUDE IN THE DESCRIPTION

This list of the documents listed by the applicant has been generated automatically and is included solely for the better information of the reader. The list is not part of the German patent or utility model application. The DPMA assumes no liability for any errors or omissions.

Cited patent literature

• US 8386188 [0010]
• US 8321150 [0010]
• US 8315814 [0010]
• US 8315813 [0010]
• US 8315812 [0010, 0037, 0101]
• US 8311750 [0010]
• US 8311748 [0010]
• US 8311747 [0010]
• US 8157742 [0010]

Claims (10)

 A system for assessing the quality of medical images of at least a portion of the anatomy of a patient, the system comprising: A digital storage device that stores instructions for assessing the quality of medical images of at least a portion of the anatomy of a patient, and A processing unit configured to execute the instructions to perform a method comprising: Receiving one or more images of at least a portion of the anatomy of a patient, Performing an anatomical localization or modeling of at least part of the anatomy of the patient based on the received images, Obtaining for a selected one of the received images a plurality of image characteristics of the selected image of the received images using the processing unit of the computer system, wherein the plurality of image properties comprises two or more different types of image properties and wherein each of the plurality of image properties is based on an anatomical peculiarity of the anatomy of the patient is connected to the anatomical localization or modeling, and Calculating an image quality score for the selected image of the received images based on the plurality of image properties of the selected image of the received images using the processing unit of the computer system.  The system of claim 1, wherein the plurality of image characteristics comprises two or more of: a local contrast level, a local noise level, a registration error, a misalignment, a local motion anomaly, a local blur anomaly, a partial volume effect, an overexposure effect, and an artifact.  The system of claim 1 or 2, wherein the processing unit is further configured to determine one or more image features of the received images, and wherein the one or more image features comprise one or more of: an image resolution; Thickness of a medical anatomical slice image, the number of slices taken, a medication parameter and a patient characteristic.  The system of claim 1, wherein the processing unit is further configured to calculate a local score with respect to each of the one or more image properties and the image score based on the local score for each of the one or more Image property (s) were calculated.  The system of claim 4, wherein the processing unit is further configured to establish one or more connections between the one or more local rating numbers and one or more of: a modeling or simulation accuracy, a modeling or simulation precision, and a modeling or simulation performance ,  A system according to any one of claims 3 to 5, wherein the one or more image features is / are a global image feature (s).  A system according to any one of claims 1 to 6, wherein the processing unit is further adapted to extract image data from the obtained images, To determine an area, criteria for acceptance / rejection or absence associated with the extracted image data, and Determine a subset of the received images for use for anatomical localization or modeling based on the region, acceptance / rejection criteria, or absence associated with the extracted image data. The system of claim 1, wherein the processing unit is further configured to generate an anatomical model of at least a portion of the anatomy of a patient based on the anatomical location or modeling, a local uncertainty value for the anatomical model based on the one or more anatomical models predicting a plurality of image property (s) and the one or more image features, - calculating a reliability interval of a simulated blood flow calculation based on the predicted local uncertainty values, and Calculate an overall uncertainty value for the anatomical model as a quantitative measure of the image quality of the one or more images.  The system of any one of claims 1 to 8, wherein the anatomy of the patient comprises at least a portion of the patient's bloodstream, and wherein the processing unit is further configured to: To generate an anatomical model of at least part of the patient's bloodstream based on the anatomical location or modeling, To generate a calculation model of a blood flow property based on the anatomical model and Determine a blood flow property within the patient's bloodstream based on the anatomical model and the blood flow characteristic calculation model of the patient.  The system of any one of claims 1 to 9, wherein the anatomy of the patient comprises at least a portion of the patient's bloodstream, and wherein the processing unit is further configured to: To generate a calculation model for a blood flow property based on the anatomical localization or modeling and To determine a location of a functional significant constriction in a blood vessel of the patient's bloodstream based on the determined blood flow characteristic.

Images